36d4aa4 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

Textile reader: modified str to handle acronyms, hyphens.

36d4aa4John MacFarlane16 years ago

src/Text/Pandoc/Readers/Textile.hs+16−3

compare with master

unifiedsplit

@@ -70,7 +70,8 @@ import Text.Pandoc.Readers.HTML ( htmlTag, htmlEndTag, -- find code blocks

70 rawHtmlBlock, rawHtmlInline )

71import Text.Pandoc.Readers.Markdown (smartPunctuation)

72import Text.ParserCombinators.Parsec

73−import Data.Char ( digitToInt )

73+import Data.Char ( digitToInt, isLetter )

74+import Control.Monad ( guard )

75-- | Parse a Textile text and return a Pandoc document.

76-- | Parse a Textile text and return a Pandoc document.

76readTextile :: ParserState -- ^ Parser state, including options for parser

77readTextile :: ParserState -- ^ Parser state, including options for parser

@@ -312,7 +313,19 @@ inlineParsers = [ autoLink

312

313

313-- | Any string

314-- | Any string

314str :: GenParser Char ParserState Inline

315str :: GenParser Char ParserState Inline

316+str = do

315−str = many1 (noneOf (specialChars ++ "\t\n ")) >>= return . Str

317+ xs <- many1 (noneOf (specialChars ++ "\t\n "))

318+ optional $ charsInBalanced '(' ')' -- drop acronym explanation

319+ -- e.g. PBS(Public Broadcasting Service)

320+ -- parse a following hyphen if followed by a letter

321+ -- (this prevents unwanted interpretation as starting a strikeout section)

322+ result <- option xs $ try $ do

323+ guard $ not . null $ xs

324+ char '-'

325+ next <- lookAhead letter

326+ guard $ isLetter (last xs) || isLetter next

327+ return $ xs ++ "-"

328+ return $ Str result

316

329

317-- | Textile allows HTML span infos, we discard them

330-- | Textile allows HTML span infos, we discard them

318htmlSpan :: GenParser Char ParserState Inline

331htmlSpan :: GenParser Char ParserState Inline

@@ -384,4 +397,4 @@ simpleInline :: GenParser Char ParserState t -- ^ surrounding parser

384 -> GenParser Char ParserState Inline -- ^ content parser (to be used repeatedly)

397 -> GenParser Char ParserState Inline -- ^ content parser (to be used repeatedly)

385simpleInline border construct = surrounded border (inlineWithAttribute) >>=

398simpleInline border construct = surrounded border (inlineWithAttribute) >>=

386 return . construct . normalizeSpaces

399 return . construct . normalizeSpaces

387− where inlineWithAttribute = (try $ optional attributes) >> inline

400+ where inlineWithAttribute = (try $ optional attributes) >> inline